Prompt library

Check whether this application is complete.

Verify:
1. Required form sections are filled.
2. Required uploads are present: %application-tasks-status(*)%
3. Motivation letter has substantive content: %application-motivation%

Analyse whether the applicant's education is sufficient for each course.

Entry requirements: %application-infoReq%
Entry assurance: %application-infoReqEntryAssurance%
First priority details: %application-infoReqEntryAssuranceInfo1stPriority%
Entry info: %application-infoReqEntry%

Identify gaps longer than 6 months in education and employment history.

Birth date: %application-profile-birth-date%
Today: %date-today%

Ignore gaps before age 16.

Draft an email to %applicant-nameFirst% about missing documents.

Today: %date-today%
Deadlines: %application-deadlines%
Programme: %application-courses%

List each missing item with its deadline.
If any deadline is within 14 days, add an urgency note.
Keep the tone helpful. Max 200 words.

Write a briefing summary for the admissions committee.

Programme: %application-courses%
Citizenship: %application-profile-citizenship%
Education: %application-education-highest-level% from %application-education-highest-institution% (%application-education-highest-country%), graduated %application-education-highest-graduation%
Programme start: %application-academicTerm-start%
Motivation: %application-motivation%

Sections: PROFILE, EDUCATION, EXPERIENCE, MOTIVATION, STRENGTHS, AREAS FOR ATTENTION.
Max 250 words.

You are an evaluator of motivation letters for a higher education programme. Output your verdict described in point 7.

You receive these inputs:
 […insert course description…]
========
- [Motivation letter] 

Your job is to:

- Detect how likely the letter is AI-generated  
- Judge how generic or specific the letter is  
- Judge how well the letter fits this particular programme  

Follow the steps below.

1. Read the context  
   - Skim the [Curriculum]  
   - Skim the [Course description]  
   - Note key modules, skills, methods, tools, and outcomes  
   - Then read the [Motivation letter] carefully

2. Detect AI-style writing  
   Look for these signals. Treat them as hints, not proof.

   Style and tone  
   - Very smooth, polished language with little emotion or personality  
   - Neutral tone with few strong opinions or honest doubts  
   - Repetitive sentence patterns and similar length sentences  
   - Heavy use of tidy transitions like “first,” “second,” “overall,” “in conclusion”  
   - Formal tone that feels like a template

   Specificity and detail  
   - Vague statements without concrete examples, dates, names, or results  
   - Few real details about projects, grades, work, or life events  
   - No mention of actual course names or modules from the [Curriculum]  
   - Reasons for choosing the programme that could apply to any programme

   Coherence and logic  
   - Almost perfect grammar across the whole text  
   - Very neat structure: intro, body, conclusion, all well balanced  
   - Long paragraphs that say little and repeat the same ideas  
   - Balanced, careful language that avoids taking any clear stance

   Repetition and patterns  
   - Repeated phrases across paragraphs  
   - Same idea restated in slightly different wording  
   - Closing paragraph that repeats the opening in a neat way

3. Look for word and phrase patterns common in AI-written motivation letters  
   These phrases alone do not prove AI use. They are only signals.  
   Pay attention if many appear together and the letter feels generic.

   Overused motivation phrases  
   - “I am passionate about…”  
   - “I have always been interested in…”  
   - “Ever since I was a child…”  
   - “I am confident that this program will…”  
   - “I strongly believe that…”  
   - “I am eager to…” / “I am excited to…”  
   - “This program will help me achieve my goals”  
   - “Pursue my academic and professional ambitions”  
   - “Make a meaningful impact”  
   - “Broaden my horizons”  
   - “Expand my skill set”  

   Over-polite or brochure-like phrases  
   - “Your esteemed university”  
   - “Your prestigious program”  
   - “World-class faculty”  
   - “State-of-the-art facilities”  
   - “Cutting-edge research”  
   - “Renowned academic environment”  

   Filler or template-style sentences  
   - “I am writing to express my interest in…”  
   - “This opportunity represents a significant step in my career”  
   - “I am fully committed to giving my best”  
   - “My background has prepared me for this challenge”  
   - “I am confident that my skills and experiences make me a strong candidate”  

   Teaching or outline phrases  
   - “First, I will explain…”  
   - “Secondly, I would like to highlight…”  
   - “Finally, I would like to conclude by…”  
   - “There are several reasons why I am applying…”  

4. Compare with typical human traits in motivation letters  
   Human letters often include:  
   - Concrete events: “During my second year, I did…”, “Last summer, I…”  
   - Specifics: course names, project titles, tools, software, datasets, grades  
   - Realistic emotions: doubts, failures, changes of direction  
   - Slight inconsistencies in style or tone  
   - Small errors, slang, or informal phrasing here and there  
   - Personal voice: unusual word choices, jokes, or very direct statements

5. Judge how generic or specific the letter is  
   Ignore AI for a moment. Focus on content quality and fit.

   Check specificity to this programme  
   - Does the letter mention concrete modules from the [Curriculum]?  
   - Does it mention skills or methods that appear in the [Course description]?  
   - Does the applicant explain why these exact courses matter to their goals?  
   - Does the letter show that the writer actually read the [Curriculum]?  

   Check generic signs  
   - Reasons that could apply to almost any programme or university  
   - Only high-level praise: “excellent education,” “high quality teaching”  
   - No mention of teaching formats, labs, projects, or assessment style  
   - No concrete link between their past experience and this specific programme  

   Rate generic vs specific on a 1–5 scale:  
   - 1 = Very generic, could be sent to any programme  
   - 2 = Mostly generic, with a few surface-level references  
   - 3 = Mixed, some specific links but also many template phrases  
   - 4 = Mostly tailored, with clear references to this programme  
   - 5 = Highly tailored, clearly written for this exact [Curriculum] and [Course]


6. Make the AI vs human judgment  
   Based on all signals, decide how likely the letter is AI-generated.

   Allowed labels:  
   - “Very likely AI-generated”  
   - “Possibly AI-generated”  
   - “Unclear”  
   - “Likely human-written”

OUTPUT INSTRUCTIONS: 

   1. AI judgment  
      - One of: “Very likely AI-generated”, “Possibly AI-generated”, “Unclear”, “Likely human-written”  

   2. Generic vs specific rating  
      - Score: 1–5  
      - One short sentence on why  

   3. Programme fit rating  
      - Score: 1–5  
      - One short sentence on why  

   4. Reasons (3–8 bullet points)  
      - Point to concrete phrases or parts of the letter  
      - Mention AI-style signals you saw  
      - Mention signs of personal detail or real experience  
      - Mention links (or lack of links) to the [Curriculum] and [Course description]

Evaluate how well the applicant's goals align with the programme.

Motivation: %application-motivation%
Applied courses: %application-courses%

[INSERT PROGRAMME DESCRIPTION HERE]

Provide feedback on the applicant's portfolio.
Applied programme: %application-courses%

Score 1–5 on: concept, originality, technical execution, narrative coherence, programme relevance.
One sentence rationale per criterion.
Then list the 3 most impactful improvements for the next 2 weeks.
300–500 words.

Summarise the recommendation letter(s) in under 120 words.

Applied programme: %application-courses-fullName(1)%

For each letter: recommender's relationship, top 2–3 qualities with evidence, any reservations, overall tone.
Flag if the letter appears generic.

Calculate the average duration of employment in days.

If start day is missing, assume the 1st.
If end day is missing, assume the last day of the month.
Today for current positions: %date-today%

Calculate the "Education Density" for the applicant. This metric identifies the percentage of time the applicant was enrolled in education, treating short gaps (e.g., summer breaks between degrees) as continuous enrollment.

Context & Rules:

Input Data: You will receive a list of education records. Each record has a start_date and an end_date.

Date Handling & Imputation:

Single Point of Truth for "Today": The reference date for all "present" calculations is %date-today%. This date MUST be used as the end_date for any record where end_date is "Present" or null.

Partial Date Formats: Input dates may be in YYYY-MM-DD, YYYY-MM, or YYYY format.

Start Date Imputation:

If start_date is YYYY-MM, assume the first day: YYYY-MM-01.

If start_date is YYYY, assume the first day of the year: YYYY-01-01.

End Date Imputation (if not "Present" or null):

If end_date is YYYY-MM, assume the last day of that month (e.g., 2020-06 -> 2020-06-30).

If end_date is YYYY, assume the last day of that year: YYYY-12-31.

Inclusivity: All date duration calculations must be inclusive (i.e., end_date - start_date + 1 day).

"Natural Gap" Definition: A "natural gap" between two education programs is 122 days or less (approx. 4 months).

Calculation Steps:

a.  Impute and Sort:
i.   For every record, determine its precise start_date and end_date using the Date Handling rules.
ii.  Sort the list of records by start_date (earliest to latest).

b.  Find Contiguous Blocks (Merge Logic):
i.   Create a list of "education blocks," starting with the first record.
ii.  Iterate through the sorted records. For each new record, compare it to the last block in your list:
* Overlap: If the new record's start_date is on or before the last block's end_date, it's an overlap. "Merge" it by updating the block's end_date to be the later of the two end dates.
* Natural Gap: Calculate the gap (in days) between the last block's end_date and the new record's start_date. If the gap is 122 days or less, "merge" it. The block's end_date is now the new record's end_date.
* Significant Gap: If the gap is more than 122 days, this is a new contiguous block. Add it to your list of blocks.
iii. You will be left with one or more "contiguous education blocks."

c.  Calculate TotalDaysInBlocks (Numerator):
i.   For each contiguous block, calculate its total duration (from its start_date to its end_date), inclusive.
ii.  Sum the durations of all blocks. This is the TotalDaysInBlocks.

d.  Calculate TotalEducationSpan (Denominator):
i.   Find the start_date of the very first block.
ii.  Find the end_date of the very last block.
iii. Calculate the total duration between these two dates, inclusive. This is the TotalEducationSpan.

e.  Handle Edge Cases:
i.   If there are no education records, the Education Density is 0.
ii.  If TotalEducationSpan is 0 (e.g., a one-day course), the Education Density is 100.

f.  Calculate Education Density:
i.   Density = (TotalDaysInBlocks / TotalEducationSpan) * 100

Output Format: Respond with only the final calculated number, rounded to two decimal places. Do not add any explanatory text, labels, or units (like '%').

Calculate the "Employment Density" for the applicant. This is the percentage of time the applicant has been employed within their total documented career span.

Context & Rules:

Input Data: You will receive a list of employment records. Each record has a start_date and an end_date.

Date Handling & Imputation:

Single Point of Truth for "Today": The reference date for all "present" calculations is %date-today%. This date MUST be used as:

The end_date for any record where end_date is "Present" or null. This overrides all other end date rules.

The end point for the TotalCareerSpan calculation.

Partial Date Formats: Input dates may be in YYYY-MM-DD, YYYY-MM, or YYYY format.

Start Date Imputation:

If start_date is YYYY-MM, assume the first day: YYYY-MM-01.

If start_date is YYYY, assume the first day of the year: YYYY-01-01.

End Date Imputation (if not "Present" or null):

If end_date is YYYY-MM, assume the last day of that month (e.g., 2020-06 -> 2020-06-30).

If end_date is YYYY, assume the last day of that year: YYYY-12-31.

Inclusivity: All date duration calculations must be inclusive (i.e., end_date - start_date + 1 day).

Calculation Steps:

a.  Calculate TotalDaysEmployed:
i.   For each employment record, find its precise start_date and end_date using the Date Handling & Imputation rules.
ii.  Calculate the total number of days for that record.
iii. Sum the durations from all records to get TotalDaysEmployed.

b.  Calculate TotalCareerSpan:
i.   Find the earliest (imputed) start_date from all employment records. This is the CareerStartDate.
ii.  Calculate the total number of days between the CareerStartDate and the PresentDate (%date-today%), inclusive. This is the TotalCareerSpan.

c.  Handle Edge Cases:
i.   If there are no job records, the Employment Density is 0.
ii.  If TotalCareerSpan is 0 (e.g., first job started today), the Employment Density is 100.

d.  Calculate Employment Density:
i.   Density = (TotalDaysEmployed / TotalCareerSpan) * 100

Output Format: Respond with only the final calculated number, rounded to two decimal places. Do not add any explanatory text, labels, or units (like '%').

Convert the applicant's grade to a percentage (0–100).

Grade: %application-education-highest-level%
Institution: %application-education-highest-institution%
Country: %application-education-highest-country%
Programme: %application-education-highest-programme-name%

Conversions: USA 4.0→(GPA/4)×100. UK: First=85, 2:1=68, 2:2=58. Germany 1–5→((5-grade)/4)×100. France /20→(grade/20)×100. India %→as-is.
Other scales: estimate position in pass-to-max range.

Calculate the number of days between graduation %application-education-highest-graduation% and programme start %application-academicTerm-start%.

Cross-reference all submitted documents for %applicant-name(uppercase)% (ID: %applicant-id%).

Secondary graduation: %application-education-last-secondary-graduated%
Birth date: %application-profile-birth-date%
Citizenship: %application-profile-citizenship%
Today: %date-today%

Check: internal consistency, timeline coherence, document formatting, translation accuracy, digital artifacts.

Compare the applicant's name across all documents.

Form name: %applicant-name%
Legal name: %applicant-nameLegal%
First: %applicant-nameFirst%
Last: %applicant-nameLast%

Check age eligibility.

Birth date: %application-profile-birth-date%
Programme start: %application-academicTerm-start%
Today: %date-today%
Education level: %application-education-highest-level%
Graduation: %application-education-highest-graduation%

Minimum age: 17 at programme start.
Flag if bachelor's holder under 20 or master's under 22.

Verify an applicant's declared education history against their attached supporting documents. You must check for two things:

1) Existence: Every declared education record has a corresponding supporting document.

2) Consistency: The start_date and end_date on the application form are consistent with the dates found on the documents.


Verification Rules & Logic:

Matching: For each record in Input A, you MUST find one corresponding document in Input B. Matching should be based on institution and degree (minor variations are acceptable, e.g., "MSc" vs "Master of Science").

Date Consistency (Internal Logic):

1) Exact Match: Declared 2017-09-01 matches Document Sept 1, 2017.

2) Partial Match (Inclusive): Declared 2023-09 matches Document Fall 2023.

3) Partial Match (Inclusive): Declared 2021 matches Document 2021-05-10.

4) Present Match: Declared end_date of "Present" is consistent if the document shows the program is "in-progress," "active," or has a future end date.

5) Mismatch (Contradiction): Declared 2017-09 would NOT match Document 2017-10-01.


6) Missing secondary and higher education documents labels



In-Progress Studies (Special Rule): For any record where end_date is "Present" or null:

If a document (like a confirmation of enrolment) is found, apply the normal verification logic.

If a document is not found, this is acceptable. Do not fail the record. Use the Pending - In Progress status.

Role: You are a Chronological Visual Analyst. Your task is to validate the logical sequencing of two images based on visual age markers.

Input Context:

Image A (Passport): A document containing an Issue Date. (Chronologically, this must represent the subject's past appearance).

Image B (Profile picutre): A current profile photograph. (Chronologically, this must represent the subject's present appearance).

Strict Constraints:

Ignore image quality (blur, B&W, watermarks).

Focus only on biological morphology.

Step-by-Step Analysis:

Timeline Extraction:

Extract the Issue Date from Passport .

Calculate the Time Delta (years passed) between the Issue Date and Today  %date-today% .

Logical Rule: Passport represents the subject [Time Delta] years ago. Therefore, Image A must appear younger (or equal) to Profile picture.

Anatomical Feature Extraction:

Compare invariant features (Bone structure: Jaw, Orbits, Ears). Are they structurally similar?
Do these two people look similar. Can this be the same person at different age?

Directional Aging Verification (Crucial Step):

Analyze the Signs of Aging in both images (e.g., skin elasticity, nasolabial folds, jowls, hairline, volume loss).

Compare the "Visual Age":

Does  (Passport) show more signs of aging (e.g., deeper wrinkles, looser skin) than Profile picture B?

Does (Profile picture) show fewer signs of aging (e.g., tighter skin, fuller face) than Passport?

Constraint: If the Document (Image A) looks biologically older than the Profile (Image B), this violates the chronological timeline.

Final Output:

Time Delta: [Number] Years.

Structural Match: [Yes/No regarding resemblance ].

Aging Direction: State "Correct (Person in the passport picture looks younger than in the profile picture)" or "Inverted (Person in passport picture appears to be older than on the profile picutre)".

Visual Logic Assessment:

Rate as "Consistent" ONLY if: Features match AND Aging direction is correct.

Rate as "Inconsistent" if: Features do not match OR if the Document photo appears biologically older than the current Profile.

Scan the full application for red flags.

Applicant: %applicant-name%
Birth date: %application-profile-birth-date%
Graduation: %application-education-highest-graduation%
Courses: %application-courses%
Motivation: %application-motivation%

Check for: document inconsistencies, timeline implausibilities, motivation letter AI-generation signals, missing documents, credential concerns.

Classify the applicant's highest education level.

Level: %application-education-highest-level%
Institution: %application-education-highest-institution%
Programme: %application-education-highest-programme-name%
Country: %application-education-highest-country%
Graduation: %application-education-highest-graduation%

Objective: Create a structured chronological timeline of documented activities for %applicant-name% (ID: %applicant-id%).

Data sources to analyze:

Birth: %application-profile-birth-date% in %application-profile-birth-country%

Current date: %date-today(locale)%

Instructions:

Extract all activities with start and end dates from provided sections.

Sort chronologically from birth to present.

Identify temporal gaps (>3 months with no documented activity).

Identify overlaps (concurrent activities in different categories).

Format using indentation levels:

Level 0: Major life phases (Childhood, Education, Professional)

Level 1: Years or year ranges

Level 2: Specific activities

Level 3: Details and gap/overlap notifications


Output format requirements:

Use consistent date format: YYYY-MM or YYYY-MM-DD

Mark gaps as: [GAP: X months - no documented activity]

Mark overlaps as: [OVERLAP: Activity A concurrent with Activity B]

Include location changes from residence history

Maximum 100 lines


Example structure: 

CHILDHOOD & EARLY EDUCATION 1995-01-15: 
Born in [Country] 2001-09: Started primary education Location: [City, Country] 2007-06: Completed primary education [GAP: 3 months - no documented activity] 2007-09: Started secondary education

EDUCATION PHASE 2013-2018: Bachelor's degree Institution: [Name] Location: [City, Country] [OVERLAP: Part-time employment concurrent with studies] 2016-2018: Part-time employment Position: [Title] Company: [Name]

PROFESSIONAL PHASE 2018-07 to 2020-03: Full-time employment Position: [Title] Company: [Name] [GAP: 6 months - no documented activity] 2020-09 to present: Current position

SUMMARY Total documented gaps: X periods totaling Y months Total documented overlaps: Z instances Timeline span: YYYY to YYYY (X years)

Map pending tasks to upcoming deadlines.

Deadlines: %application-deadlines%
Term start: %application-academicTerm-start%
Task statuses: %application-tasks-status%

List the nearest 5 dates with their dependent tasks.

Generate an email with a prioritised action list for %applicant-name%.

Today: %date-today%
Deadlines: %application-deadlines%
Term start: %application-academicTerm-start%
Courses: %application-courses%

Include up to 10 actions, each with an ISO deadline and brief rationale.

Compare the submitted diploma with standard formats for the applicant's country.

Country: %application-education-last-secondary-country%
Institution: %application-education-last-secondary-institution%
Citizenship: %application-profile-citizenship%

Check: font style, seal placement, signatures, security features, layout.

Check submitted materials against conditional requirements.

Today: %date-today%
Programme start: %application-academicTerm-start%
Task statuses: %application-tasks-status%

Requirements:
[INSERT REQUIREMENTS HERE]

Extract language test results from certificates and the form.

English requirement: %application-englishRequirementsHighest%

For each test: name, overall score, sub-scores if available, date (YYYY-MM-DD).
Most recent only per test type.

Extract research experience from the CV and any research statement.

For each: title, type (publication/thesis/conference/poster/other), venue, year, role.
Max 10 entries (most recent). Provide counts at the end.